[00:39] <antarus> Does the installer support GPT ?
[00:40] <cjwatson> Yes
[00:44] <antarus> I'm not super familiar with GPT
[00:44] <antarus> but it appears that it is not being configured correctly, grub is telling me "This GPT partition label has no BIOS Boot Partition; embedding won't be possible!."
[00:45] <antarus> then later on /usr/sbin/grub-setup: error: cannot read `/grub/core.img' correctly.
[00:45] <cjwatson> Are you using manual partitioning?  How old an image are you using?
[00:45] <antarus> this bug was filed march 23, we sync images about 12h after you release them
[00:45] <antarus> let me double check the preseed
[00:45] <cjwatson> That's odd because I added a big warning to the partitioner before that that complains if you don't have such a partition.
[00:46] <cjwatson> Although if you're preseeding a partitioning layout then it's your job to create such a partition.
[00:46] <antarus> we are using partman_auto/expert_recipe
[00:46] <antarus> (it was like that when I got here, etc..)
[00:46] <antarus> so we need to change our preseed then yes?
[00:47] <cjwatson> 1 1 1 free $iflabel{ gpt } $reusemethod{ } method{ biosgrub } .
[00:47] <antarus> excellent
[00:47] <cjwatson> is the bit from the default atomic recipe
[00:47] <cjwatson> I'd probably put it at the start
[00:47] <antarus> as always your assistance was quick and greatly appreciated ;)
[00:47] <cjwatson> thank me when it works ;-)
[00:48] <antarus> yes yes ;p
[00:59] <antarus> hrm, any idea if this will work fine on non-gpt as well?
[01:30] <cjwatson> antarus: yes, the $iflabel{ gpt } bit arranges for this to be a no-op elsewhere
[01:38] <antarus> ahhh
[02:56] <CIA-32> debian-installer: cjwatson * r1334 lucid-proposed/ (8 files in 2 dirs): Move to 2.6.32-40 kernels.
[02:58] <CIA-32> debian-installer: cjwatson * r1335 lucid-proposed/ (9 files in 3 dirs): Move oneiric-* images to 3.0.0-17 kernels.
[02:59] <CIA-32> debian-installer: cjwatson * r1336 lucid-proposed/debian/changelog: releasing version 20081029ubuntu102.15
[02:59] <antarus> oh dear lord, did we build our own partitioner?
[02:59]  * antarus ughs
[02:59] <antarus> the past, I hate it
[03:01] <cjwatson> this is why we went to enormous effort to reuse d-i's partitioner in ubiquity, despite the complexity induced by doing so
[03:01] <cjwatson> maintaining two => ftl
[03:03] <antarus> I love it when my machine is misconfigured to not print to console, and my syslogs stopped 30 minutes ago with no errors ;p
[03:08] <antarus> Used to install additional, non-standard d-i components (udebs). It will check if anna has already been run. If it has, the component is unpacked immediately; if it has not, it will be scheduled for installation when anna is run.
[03:08] <antarus> aha
[03:09] <antarus> I wonder if anna moved around in d-i between L and P
[03:12] <antarus> cjwatson: is there a document detailing the steps of the installer?
[03:13] <antarus> ahh I guess main-menu is dynamically assigned..
[03:13] <antarus> bleh
[03:16] <cjwatson> antarus: http://d-i.alioth.debian.org/doc/internals/
[03:17] <cjwatson> not totally up to date but pretty good
[03:17] <cjwatson> I don't believe anna moved
[03:17] <antarus> it wouldn't surprise me if this never worked either ;p
[03:18] <antarus> I guess the quickest thing to do is take the initrd and just bust it open
[03:39] <antarus> oh look
[03:39] <antarus> I pulled some lucid logs and it never worked there either
[03:39]  * antarus sighs
[10:16] <gema> does anyone know where I can have a look at the code of ubiquity_webcam_play ?
[10:18] <cjwatson> gema: lp:ubiquity, src/webcam/webcam.c
[10:19] <gema> cjwatson: thanks
[10:30] <gema> cjwatson: we've set ulimit -c unlimited but after sending SIGILL or SIGSEV there is no core dump from ubiquity on the live install, is there anything we are missing?
[10:31] <gema> cjwatson: I am trying to leave enough info on bug 966294 so that you guys can actually fix it, it has to do with the camera
[10:31] <ubot2> Launchpad bug 966294 in ubiquity "Precise AMD64 Desktop beta 2 image hangs during an install" [Undecided,New] https://launchpad.net/bugs/966294
[10:31] <cjwatson> It might have a current directory it can't write to.
[10:31] <cjwatson> However surely the crash reporting thing should get hold of the core dump regardless.
[10:32] <gema> cjwatson: we are running it manually from ~ , does it change its current directory?
[10:32] <gema> cjwatson: we are trying to make it coredump with a signal
[10:32] <cjwatson> Not explicitly
[10:33] <cjwatson> Anything in /var/crash afterwards?
[10:33] <gema> no
[10:34] <cjwatson> Not sure then :-(
[10:34] <cjwatson> You could attach gdb without crashing the process
[10:34] <cjwatson> 'gdb /usr/bin/python process-id-of-ubiquity'
[10:34] <cjwatson> maybe sudo
[10:34] <davmor2> cjwatson, gema: is it stopping once you click on the continue button post user details?
[10:34] <gema> we already attached it, got thread info and stack trace of the relevant thread
[10:34] <gema> however without symbols that's not all that helpful
[10:35] <gema> so we are trying to get a core dump to debug it on a machine with the symbols
[10:35] <gema> davmor2: yes
[10:35] <cjwatson> Or you could grab the symbols from ddebs.ubuntu.com first
[10:35] <cjwatson> And install them in the live session
[10:35] <cjwatson> That would be my approach
[10:36] <gema> cjwatson: ack
[10:36] <davmor2> gema: I got it here I worked through all the files I could with evan and we couldn't find anything useful either
[10:36] <gema> davmor2: the camera setup is failing
[10:36] <gema> davmor2: we are trying to get some debuggable stuff out of it
[10:37] <gema> davmor2: we've narrowed it down to ubiquity_webcam_play with gdb
[10:37] <cjwatson> That function itself, or the gst stuff it calls?
[10:37] <cjwatson> That function is really very simple indeed
[10:38] <cjwatson> void
[10:38] <cjwatson> ubiquity_webcam_play (UbiquityWebcam *webcam) {
[10:38] <cjwatson>         UbiquityWebcamPrivate *priv = UBIQUITY_WEBCAM_PRIVATE (webcam);
[10:38] <gema> cjwatson: the gst stuff, would be our guess
[10:38] <cjwatson>         if (gst_element_set_state (priv->camerabin, GST_STATE_PLAYING) == GST_STATE_CHANGE_FAILURE) {
[10:38] <cjwatson>                 g_print ("setting camerabin to PLAYING failed\n");
[10:38] <cjwatson>                 return;
[10:38] <cjwatson>         }
[10:38] <cjwatson> }
[10:39] <davmor2> gema: yeah that's what I had a chat with ev about while I was down in millbank
[10:41] <CIA-32> ubiquity: cjwatson * r5343 trunk/ (debian/changelog src/webcam/webcam.c):
[10:41] <CIA-32> ubiquity: Exit early from ubiquity_webcam_play and ubiquity_webcam_stop if priv or
[10:41] <CIA-32> ubiquity: priv->camerabin is NULL.
[10:41] <cjwatson> ^- I don't *think* that's your bug but I happened to notice it
[10:42] <gema> ours gets stuck there forever, until it runs out of memory
[10:51] <cjwatson> do you have any of the backtrace at all?
[10:51] <cjwatson> even without symbols it might be worth a brief look
[11:33] <gema> cjwatson: I attached all the info I could gather, one file has most symbols loaded, bug 966294
[11:33] <ubot2> Launchpad bug 966294 in ubiquity "Ubiquity loops forever from ubiquity_webcam_play" [Undecided,New] https://launchpad.net/bugs/966294
[11:33] <gema> cjwatson: hope it helps, because it was a painful process :)
[11:35] <cjwatson> Do you still have that gdb running?
[11:35] <gema> cjwatson: yes
[11:36] <cjwatson> Could you go up to frame 35 (ubiquity_webcam_play), then 'p priv' and 'p priv->camerabin'?
[11:36] <gema> priv is optimized out
[11:36] <cjwatson> Oh, hah, that's at the bottom
[11:36] <cjwatson> OK, not NULL at least
[11:36] <gema> that's the closest we could get
[11:37] <cjwatson> That's an odd place to hang.  Does it sit there?
[11:37] <cjwatson> Oh, I suppose if it's copying something unbounded.
[11:37] <gema> it just keeps going until it runs out of memory at some point, the UI unresponsive
[11:42] <cjwatson> It smells of a non-NULL-terminated list or something.
[11:47] <cjwatson> This will probably take ages to run, and I'm not entirely certain it won't break something else, but could you try running ubiquity under 'valgrind --log-file=ubiquity.vg --error-limit=no --num-callers=50'?
[11:48] <cjwatson> I can try zenning it from the code in parallel, but that's a rather uncertain process.
[11:58] <cjwatson> Wonder if there's any hope of reproducing this in the test suite
[12:16] <cjwatson> Hm, valgrind doesn't seem to have a problem with this in the test suite for me.
[12:26] <gema> cjwatson: I was cooking, reading you now
[12:50] <gema> cjwatson: running, it is taking a while, though
[13:20] <gema> cjwatson: with trace-children, valgrind exits before ubiquity actually starts and without trace-children it doesn't trace the right process
[13:21] <gema> cjwatson: not sure what else to do, I could ship the netbook to you, for debugging purposes?
[13:22] <cjwatson> See previous comments about lack of space
[13:22] <gema> cjwatson: you can return it to me when you are done!
[13:22] <cjwatson> Hmm
[13:22] <gema> cjwatson: it is a small pink testing netbook
[13:23] <cjwatson> Well, there's that.  Dunno.  I guess maybe, if it won't leave you short.  Will you be at the release sprint?
[13:23] <cjwatson> (It'd be too late to fix it then, but I could return it then, as an option.)
[13:23] <gema> cjwatson: I wasn't planning to be, but I can come if necessary, if it is only for returning it, you can give it to michelle and she'll send it to me
[13:26] <cjwatson> I think I should see if it reproduces on my own laptop first, though I really ought to write my performance review before rebooting for that :-)
[13:26] <gema> cjwatson: ack, I am trying something else with jibel meanwhile
[13:26] <gema> cjwatson: we can arrange for this after beta anyway
[14:39] <gema> davmor2: you said you were discussing this ubiquity hang problem with ev back in london
[14:39] <gema> davmor2: can you check with camera do you have?
[14:44] <davmor2> gema: http://davmor2.co.uk/pc-specs/main-laptop.html
[14:46] <gema> davmor2: I don't find it there
[14:47] <davmor2> gema: meh give me 5
[14:47] <gema> davmor2: no hurry
[14:48] <gema> davmor2: lsusb should tell you
[14:49] <davmor2> gema: http://paste.ubuntu.com/904002/ output of lsusb -v
[14:53] <gema> davmor2: ta
[15:31] <gema> cjwatson: we finally managed to get a successful run of valgrind with memcheck option, which led us to decide to also gather an strace for open and ioctl
[15:31] <gema> cjwatson: logs attached to bug 966294
[15:31] <ubot2> Launchpad bug 966294 in ubiquity "Ubiquity loops forever from ubiquity_webcam_play" [Undecided,New] https://launchpad.net/bugs/966294
[16:04] <CIA-32> apt-setup: cjwatson * r230 ubuntu/ (debian/changelog generators/50mirror.ubuntu): generators/50mirrors.ubuntu: do not reuse the file variable (LP: #964930)
[16:04] <cjwatson> ^- from bdmurray
[16:04] <cjwatson> gema: thanks
[16:04] <gema> np
[16:06] <CIA-32> apt-setup: cjwatson * r231 ubuntu/debian/changelog: releasing version 1:0.55ubuntu2
[16:09] <antarus> Ahh I forgot that cjwatson never sleeps ;)
[16:09] <infinity> It's not that late in Cambridge...
[16:10] <cjwatson> Didn't sleep much last night, mind.  Zoning out a bit here.
[16:10] <infinity> I slept an hour last night.
[16:10] <infinity> Which happened to overlap 11 minutes into our morning meeting.
[16:10] <infinity> So, go me.
[16:10] <infinity> \o/
[20:17] <antarus> ohh d-i
[20:17] <antarus> how I hate you
[22:20] <antarus> cjwatson: hmm, can't use "d-i partman-auto/method string lvm" with gpt ?
[22:20] <antarus> or I'm just doing this recipe wrong
[22:20]  * antarus goes back to looking at partman
[22:40] <antarus> cjwatson: as usual, bug in my preseed